Doped Amorphous Ti Oxides To Deoptimize ... - ACS Publications

Subscriber access provided by NEW YORK UNIV

Article

Doped Amorphous Ti Oxides to Deoptimize Oxygen Reduction Reaction Catalysis Mitchell C. Groenenboom, Rachel M. Anderson, Derek J. Horton, Yasemin Basdogan, Donald F. Roeper, Steven A. Policastro, and John A. Keith J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.7b04210 • Publication Date (Web): 03 Jul 2017 Downloaded from http://pubs.acs.org on July 4, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry C is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Doped Amorphous Ti Oxides to Deoptimize Oxygen Reduction Reaction Catalysis Mitchell C. Groenenboom,1 Rachel M. Anderson,2 Derek J. Horton,3 Yasemin Basdogan,1 Donald F. Roeper,4 Steven A. Policastro,3,* John A. Keith1,* 1

Department of Chemical and Petroleum Engineering, University of Pittsburgh, Pennsylvania

15261, United States 2

National Research Council Postdoctoral Associate at the U.S. Naval Research Laboratory,

Washington, DC 20375, United States 3

Center for Corrosion Science and Engineering, U.S. Naval Research Laboratory, 4555

Overlook Ave., SW Washington, DC 20375, United States 4

Excet, Inc., Springfield, Virginia 22151, United States

ABSTRACT: The oxygen reduction reaction (ORR) is a major factor that drives galvanic corrosion. To better understand how to tune materials to better inhibit catalytic ORR, we have identified an in silico procedure for predicting elemental dopants that would cause common, natively formed titanium oxides to better suppress this reaction. In this work, we created an amorphous TiO2 surface model that is in good agreement with experimental radial distribution function data and contains reaction sites capable of replicating experimental ORR overpotentials. Dopant performance trends predicted with our quantum chemistry model mirrored experimental

ACS Paragon Plus Environment

1


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 21

results, and our top three predicted dopants (Mn, Al, and V, each present at doping concentrations of 1%) were experimentally verified to lower ORR currents under alkaline conditions by up to 77% vs. the undoped material. These results show the robustness of calculated thermodynamic descriptors for identifying poor, TiO2 based ORR catalysts. This also opens the possibility of using quantum chemistry to guide the design of coating materials that would better resist the ORR and presumably galvanic corrosion.

Introduction Worldwide efforts to prevent and remediate corrosion damage cost approximately 3.4% of the global GDP per year.1 Even materials that are normally resistant to corrosion in isolation can degrade due to galvanic couples that form between dissimilar metals. This effect is powerful enough to corrode aircraft grade aluminum alloys despite being covered by their stable, corrosion resistant metal oxides.2-3 Problems such as these have inspired many efforts to create better functional coatings that physically block corrosive conditions,4-5 sacrificially corrode to protect the substrate,6 or kinetically slow corrosion rates.7-8 Galvanic corrosion occurs near boundaries between dissimilar metals (see Figure 1). Metal oxidation on the less noble metal surface (anodic site) provides electrons that drive the oxygen reduction reaction (ORR) on the more noble metal surface (cathodic site). Because the driving force for metal oxidation decreases if these electrons are not consumed, the ORR is a major factor that controls the overall galvanic corrosion rate in atmospheric environments.9-10 Thus, inhibiting catalytic ORR activity at the cathodic site is an opportunity to decrease the rate of galvanic corrosion.


2

Page 3 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. Illustration of galvanic corrosion in an atmospheric environment. The junction of the two plates of metal 1 with a noble metal fastener establishes the galvanic junction of dissimilar metals that can cause corrosion once a droplet of water forms on the surface. The high surface area-to-volume ratio of the droplet allows a high dissolved oxygen concentration even once the reduction reaction begins consuming oxygen.

Amorphous Ti oxides are an ideal candidate material to reduce cathodic kinetics because fasteners made from Ti alloys commonly form galvanic couples with other metals, such as Al and its alloys. These amorphous Ti oxides can be doped with other metals to further alter cathodic kinetics. Doped Ti oxides are reasonably stable and can be natively formed on doped Ti metal surfaces without the need for traditional barrier coatings. Moreover, the dopants can be alloyed in the Ti bulk material so the doped Ti oxide will spontaneously regenerate if damaged. Previous efforts to use TiO2 to inhibit galvanic corrosion have attempted to provide cathodic protection via photo-generated electrons from TiO2 photoanodes.11 In contrast to those works, our study focuses on identifying dopants that decrease ORR electrokinetics on amorphous Ti oxide surfaces. Computational quantum chemistry studies frequently use thermodynamic descriptors and Sabatier volcano curves to identify optimal catalysts that lie near the top of the activity volcano


3


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 21

(see for example the large body of fuel cell literature12-16). While this level of modeling works well in predicting dopants that maximize the catalytic activity of a material, these in silico models are not normally used to predict dopants that minimize catalytic activity, i.e. dopants that lie near the bottom of the Sabatier volcano plots. We report an integrated computational and experimental study that demonstrates that simple Sabatier volcano descriptors can be used to qualitatively predict metal dopants that experimentally decrease ORR currents by as much as 77% when impregnated in amorphous TiO2 at doping concentrations of 1%. This report focuses on using modeling and experiment to understand ORR catalysis on doped Ti oxide surfaces. The overall effectiveness of these coatings at reducing galvanic corrosion will be addressed in future work.

Computational Methods Unless otherwise specified, all presented energies and structures were determined using KohnSham density functional theory (DFT) in the Vienna ab initio simulation package (VASP)17-19 utilizing the Perdew-Burke-Ernzerhof (PBE)20 GGA exchange correlation functional and the projector augmented wave (PAW) method21 with spin polarization. Planewave energy cutoffs of 450 eV and a 2x2x1 k-point grid gave well-converged energies. We approximate the zero point energy, entropic, and solvation free energy of water by using the values predicted by Valdés et. al. for the ORR intermediates adsorbed to TiO2.22 When noted, VASPsol was used to account for solvated reaction energetics,23-24 using the relative permittivity of water (78.4) along with the previously mentioned parameters (PBE, PAW potentials, spin polarization enabled, 450 eV energy cutoff, and 2x2x1 k-point grid). Additional energy calculations using the HSE06 hybrid DFT functional25 (PAW method, spin polarization enabled, 450 eV energy cutoff, and a gamma-


4

Page 5 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


point calculation) were performed on PBE optimized structures to determine the impact that higher levels of theory have on intermediates adsorbed to doped amorphous Ti oxides. HSE06 calculations were performed for dopant systems where we also obtained experimental data. We created amorphous TiO2 surfaces by annealing crystalline TiO2 slabs using the reax/c26 implementation of ReaxFF27 in LAMMPS.28 All annealing simulations used the forcefield parameterized by Kim and Kubicki.29 Rutile TiO2 slabs were heated from 0 K to 1100 K at a rate of 0.06 K/fs using a 1 fs timestep. After 300 ps at 1100 K, we quenched the structures to 0 K at a rate of 0.05 K/fs. The resulting structures were then fully relaxed using DFT in VASP as described previously (PBE, PAW potentials, spin polarization enabled, 450 eV energy cutoff, and 2x2x1 k-point grid).

Experimental Methods Titanium-doped alloys were arc-melted using high purity metals (greater than 99.995 at%). Ingots were subsequently suction cast into a cylindrical copper mold. The cylinder rods were then machined to 1.1 cm and then ground to a final dimension of 1 cm. After casting, a four-hour solution anneal in argon was performed at 827°C (Ag, Al, Cr, Sn, and Ti), 685°C (Co), or 550°C (Mn), within the single-phase HCP region, followed by a water quench. The crystal structure was determined using Bragg-Brentano X-ray diffraction (XRD), with a Cu k-α source of wavelength 1.5405 Å. XRD and XPS results are summarized in Figures S1, S2, and table S1 in the supporting information. Samples of 1 cm diameter were then mounted in insulating epoxy and the surface areas were measured. Prior to electrochemical testing, samples were abraded using successively finer grits to 1200 grit SiC paper and then polished using 1 µm alumina and sonicated in water. Because titanium in its pure state is highly reactive with oxygen, the native


5


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 21

oxide was assumed to begin formation almost immediately upon completion of the polishing step. No additional anodic polarization was used to drive the oxide formation. ORR activities on the oxides were then evaluated using traditional cathodic polarization scans in 0.6 M NaCl adjusted to pH 12 with NaOH. These conditions should represent those near the cathodic material in a galvanic couple under a water droplet in atmospheric conditions.30 After an 18-hour open circuit (OC) hold, the potential was scanned in the negative direction from +0.02 V above EOC to -2.0 VSCE at 0.167 mV/s using a graphite counter electrode. Each experiment was performed in triplicate. An 18-hour OC hold showed less scatter than the 1-hour hold used in previous work,31 and therefore we have higher confidence in the values presented here.

Results and Discussion Amorphous TiO2 model. We used an atomistic reactive forcefield (ReaxFF29) to create an amorphous oxide surface model as has been done by others.32 A crystalline TiO2 surface composed of 3x3 rutile unit cells (3 tri-layers thick) was annealed using ReaxFF as described in the computational methods section. As shown in Figure 2, radial distribution functions for the slab structures agree well with experimental data for amorphous TiO2 nanoparticles after full optimization using DFT. The ReaxFF annealed structure for our system itself did not agree with experimental data,33 but this is likely due to the forcefield being parameterized to best model crystalline TiO2 structures (opposed to amorphous TiO2 structures). Although our system has a distinct unit cell and therefore is not truly amorphous, the agreement with experimental data displayed in Figure 2 shows that our system is a reasonable model for a structurally relaxed facet on an amorphous TiO2 surface where ORR could be expected to occur.


6

Page 7 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 2. The radial distribution functions for the Ti-Ti, Ti-O, and O-O pairs from the ReaxFF annealed structure, the annealed structure after being optimized with density functional theory in VASP, and experimental data from reference 33. The QM optimized structure agrees with experimental data.

Calculating ORR overpotentials. Calculating reaction overpotentials with the computational hydrogen electrode model12 is normally the first step toward modeling electrocatalytic activity. This model can yield robust insight into electrocatalytic reaction rate trends despite not explicitly calculating reaction barriers or accounting for other factors such as defects. We used this


7


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 21

approach to calculate reaction overpotentials for the associative ORR mechanism commonly used to describe the ORR on metal oxides as shown in Figure 3a.34-35 Because the hydrogen evolution reaction (1/2 H2 ⇌ H+ + e–) is in equilibrium at 0 V vs. the standard hydrogen electrode (VSHE), the energies of protons and electrons in electrochemical reduction steps were modeled as half the energy of an H2 molecule plus a linear energy correction to account for an applied potential. Using these energy corrections, we calculated the theoretical reaction overpotential by finding the applied potential at which all four reaction steps are downhill in energy. Mathematically, this was determined by the most uphill reaction step at the equilibrium potential for the ORR (1.23 VSHE). This approach assumes that the activation barrier for the rate limiting step will be at least equal to the most uphill reaction step, and this assumption has been used to successfully study electrochemical reactions (including the ORR) on metals and metal oxides as shown in references 9-13. Our amorphous TiO2 surface model contained four unique surface sites (each at a fourcoordinate surface Ti atom) on which ORR steps were considered to take place. Modeling the ORR energies on the sites shown in Figure 3b yielded overpotentials that vary by nearly 0.8 V, but the most active site (Site 1) had predicted overpotentials calculated with our PBE (without solvation) and HSE06 (with solvation) models that were in good agreement with the

experimental overpotential for TiO2 ( = 0.5 , = 0.43 , and . =

0.45 ).36 The local coordination environments at active sites has been used to describe catalytic activities.37-38 However, changes in local coordination environment were quite subtle in these amorphous structures and not pursued as a descriptor for catalysis. The agreement between previously measured ORR overpotentials and those calculated here validated that our amorphous TiO2 model can be used to study ORR mechanisms.


8

Page 9 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3. a) The associative ORR mechanism modeled in this work. ‘*’ denotes an empty surface site on the material. b) ORR reaction energies calculated with PBE for four different surface sites in the undoped TiO2 surface (labeled 1-4) plotted at an applied potential of 0 and 1.23 VSHE. The intermediates correspond to the reactions in Figure 3a.

Dopant screening. We considered different metal dopants that could be introduced into the amorphous oxide surface to increase ORR overpotentials. Each dopant atom was embedded into the surface at its preferred oxidation state given by experimental Pourbaix diagrams39 at our experimental operating conditions, –0.8 V vs the saturated calomel electrode (VSCE) at pH 12 (See Figure S4 and Table S2 for further discussion). We compared the stability of each dopant at all four different active sites as shown in Figure 3b to identify the most stable substitution site (Figures S7, S8, S9, S10, and S11 show the surface models). We assumed that the most thermodynamically stable site would reflect the atomic configuration that would be least likely to


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 21

reconstruct and would therefore have the largest effect on electrochemical ORR rates. Following work by Carter and co-workers,14 we then predicted the maximum impact of each dopant on the ORR activity by modeling the ORR intermediates adsorbed directly to the dopant atom embedded in the amorphous surface at its most stable site. The predicted maximum overpotential for each metal dopant is displayed in the Sabatier volcano diagrams shown in Figure 4. Unlike studies on fuel cell catalysis where the ideal catalyst is found at the top of the activity volcano, we sought to identify dopants at the bottom of the volcano that optimally reduce ORR rates. Our rationale for this now follows. When a reaction site near a dopant has a higher ORR overpotential than the undoped surface, that reaction site is less likely to contribute to the overall activity. Furthermore, it is highly unlikely that all reaction sites will be at a dopant because of their relatively low concentrations in the oxide (0.5-6.2%). However, dopants with high reaction overpotentials will still affect adsorbate binding energies multiple sites away,40 and the dopant’s influence on the overpotential will decrease the further the dopant is away from the reaction site. For example, the ORR overpotential at the most active site on undoped TiO2 (site 1 in Figure 3) is 0.50 V. The same site with the Al3+ dopant one nearest neighbor site away increases the overpotential at this site to 0.65 V. We note that the maximum calculated overpotential due to Al3+ is 1.43 V. Thus, the effect of the dopant on ORR overpotentials reflects the upper limit to what would be experimentally observed, but the overall trend for how dopants affect ORR activity will still be reflected by the activity volcano based on maximum dopant impact.


10

Page 11 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4. Sabatier volcano plots of computationally predicted dopant overpotentials. Dopants that were predicted and tested in this work are labeled in red, and dopants not yet experimentally verified are labeled in black. a) overpotentials calculated with PBE without solvation effects, b) overpotentials calculated with HSE06 and including solvation energies. HSE06 calculations were only performed for dopants with available experimental data. The effect of solvation is discussed further in Figure S6. We use a relatively simple surface model to screen dopants by their maximum potential to inhibit ORR activity. The model assumes that dopants at low concentrations will be distributed throughout a Ti oxide. For the cases considered here, experimental XPS measurements show


11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 21

that most dopants are present in the oxide at concentrations comparable to the original alloys (only Cr is present at higher than 2.2%, see Table S1) and so chances of segregation would be considered low. This may not always be the case when modeling complex oxides, especially those with higher dopant concentrations, and segregation may occur.41 In the best case, our model would overestimate the impact of the dopants in a segregated oxide because segregation would result in more uninhibited Ti active sites, but the trends predicted by our model would still hold. In the worst case, segregation would change the overall structure of the material which may unpredictably affect overall trends. We now discuss dopants that we verified experimentally using potentiodynamic polarization in this work (Ag, Sn, Cr, Co, Al, Mn, and V) or have been previously studied (Nb) shown in Figure 4a.36 Our computational modeling predicted that Mn and Al would bring the highest ORR overpotentials in the cases considered and thus would be the best dopants for suppressing ORR activity. Co, Sn, and Cr in turn should be moderate ORR inhibitors, and Nb and Ag should increase ORR activity relative to pure amorphous TiO2. This is consistent with previous work by Arashi and coworkers who showed that doping amorphous TiO2 with Nb slightly lowers the ORR overpotential.36 Vanadium is more challenging to characterize because it has two stable oxidations states near our experimental conditions (V3+ and V5+). Thus, V dopants are likely present as a mixture of V3+ and V5+. At more negative applied potentials, the ratio of V3+/V5+ should favor V3+ and the ORR activity of the oxide should decrease. This suggests that dopants could have different capacities to suppress ORR rates at different applied potentials in our experiments. The Pourbaix diagrams for all other considered dopants have only one stable oxidation state near our experimental conditions. We experimentally verified trends for these


12

Page 13 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


eight systems, but we also computationally predicted that Ga, Zn, and Si could be even better ORR inhibitors than Al and Mn. The Supporting Information (Figure S6) reports how Sabatier volcano diagrams are influenced using PBE/HSE06 and/or VASPsol. Figure 4b shows that using a higher level of theory (HSE06) while also accounting for solvation (VASPsol continuum solvation) does not appear to significantly alter the model predictions from PBE without solvation in most cases studied here. While the overpotentials of most of the dopants shift from those shown in Figure 4a, only Mn2+ (∆ηORR= 0.22 V), Cr3+ (∆ηORR= 0.57 V), Co2+ (∆ηORR= 0.4 V), and V3+ (∆ηORR= 0.52 V) change by more than 0.15 V. Even considering these overpotential changes the relative dopant ordering is generally preserved. The only exception to this are the predicted overpotentials of Cr3+ and Co2+. In these cases, HSE06 with VASPsol predicts that Cr3+ will be relatively less effective at inhibiting corrosion, while Co2+ should be more effective than originally predicted in Figure 4a.

Cathodic Polarization Scans. Figure 5 shows representative cathodic polarization scans for all alloys that were investigated experimentally. Note that the polarization scans were started at 20 mV above the OC potential of the alloy and scanned in the electronegative direction at a rate of 0.167 mV/s. The current density values at –0.8 VSCE, (the galvanic corrosion potential between Ti and Al alloys),30 were measured in triplicate and then averaged. The percent change for each alloy with respect to the undoped Ti are shown in Table 1. At potentials more positive than -0.8 V vs SCE, the trend for ORR suppression is consistent with what was observed at -0.8 V vs SCE – with the exception of the catalysis of the ORR on the Ti-Al and Ti-Mn alloys. The tafel slope for the ORR on both of these oxides changes around 0.65 V vs SCE which suggests that there was a change in oxidation state of the dopant atoms as


13


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 21

the potential decreased. For example, as seen in Figure 4a, V3+ and V5+ differ significantly in predicted overpotential values. As the applied potential decreases, the ratio of V3+/V5+ in the material should increase resulting in an oxide with a higher average overpotential.

Figure 5. Cathodic polarizations scans of the undoped titanium and the 1 at% doped titanium samples in air-saturated 0.6 M NaCl at pH 12 with a scan rate of 0.167 mV/s. Each scan began after an 18-hour OC hold.

Table 1. Percent change and standard deviation in current at –0.8 VSCE of alloy samples versus the undoped Ti. Alloy

Current Density (µA/cm2)

Percent Change

Ti-Ag1

17.3

+95 ± 13

Ti

8.8

–

Ti-Sn1

6.9

–21 ± 11

Ti-Cr1

6.2

–30 ± 8

Ti-Co1

4.7

–47 ± 5

Ti-Al1

3.4

–61 ± 11

Ti-Mn1

3.1

–65 ± 4

Ti-V1

2.0

–77 ± 3


14

Page 15 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Comparison to Experiment. Our quantum chemistry predictions almost exactly mirror our experimental results. The trend in dopant performance predicted by computational modeling were: Ag > undoped > Sn > Co > Cr > Al ≥ Mn >V3+ (Figure 4a) and Ag > undoped > Sn ≈ Cr > Al ≈ Mn ≈ Co > V3+ (Figure 4b) while experimental potentiodynamic polarization measurements found almost the same ranking: Ag > undoped > Sn ≥ Cr > Co > Al ≥ Mn >V Overpotentials calculated with PBE appear to overestimate the overpotential of Cr relative to Co, but these fall quite close on our volcano plot within 0.2 V (Figure 4a). While calculating the overpotentials with HSE06 and VASPsol appears to correct the relative ordering of Co and Cr, these predictions now appear to overstate the effectiveness of Co relative to Al, Mn, and V3+. The trends observed for the remaining dopants (Ag, the undoped system, Sn, Al, Mn, and V3+) are consistent between both computational models and the experimental results. Ag was anticipated to enhance electrocatalysis overall, and while the measured current increase for the Ti-Ag oxide at -0.8 VSCE agrees with our predictions, Ag exhibits anomalous polarization behavior. This may be due to Ag catalyzing additional reactions or Ag+ being reduced under operating conditions, but further analysis is outside the scope of the current study. While we exclude it from our experimental trend, our model was also in good agreement with prior studies of the ORR on Nb doped amorphous TiO2 as previously stated. The other measured current trends are consistent with the OC potential ordering until the slope of the Ti-Al and Ti-Mn alloys change relative to the other materials at ~ –0.65 VSCE causing the Ti-V and Ti-Al/Ti-Mn alloys to switch order. This may be due to the onset of different ORR


15


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 21

mechanisms on the Ti-Al and Ti-Mn alloys, or the Ti-V alloy may become less active as the V3+:V5+ ratio increases. At potentials more negative than -0.9 VSCE, the dopants could be reduced to other oxidation states not accounted for in our computational model. Still, this shows that simple Sabatier analyses are robust enough to identify dopants for materials having low catalytic activities and can be used to aid the design of protective surface treatments. The ability of the dopant atoms to bind the ORR intermediates was hypothesized to correlate with the total charge of each intermediate after it is bound to the surface. Bader charge analysis shows that *OOH bound to Al3+, Ti2+, or V3+ has a charge of –0.70, –0.97, or –1.05, respectively. Dopants with larger degrees of charge transfer bind the intermediates more tightly, effectively poisoning the surface, and they are limited by reaction 4 in Figure 3a. On the other hand, dopants that transfer less charge form weaker bonds that are less likely to form reaction intermediates and are limited by reaction 1 in Figure 3a. For dopants at intermediate oxidation states (i.e. V3+, Mn2+, Cr3+, Co2+, and Ag+), the ability to donate electron density appears to correlate with their approximate redox potentials from Pourbaix diagrams in the literature (V2O3 ⇌ V3O5 at E0 = –0.5 VSHE, Mn2+ ⇌ Mn2O3 at E0 = 0.3 VSHE, Cr2O3 ⇌ CrO42– at E0 = 0.2 VSHE, CoO ⇌ Co3O4 at E0 = 1.0 VSHE).39 For dopants at their highest oxidation state, (i.e. Nb5+, Ti4+, Sn4+, and Al3+) the ability to bind to ORR intermediates correlates with their calculated atomic radii (Nb = 1.98 Å, Ti = 1.76 Å, Sn= 1.45 Å, and Al = 1.18 Å). Although this trend might be coincidental, bonding orbitals in smaller dopants (such as Sn and Al) have less orbital overlap which makes it more difficult to transfer electron density to the adsorbed intermediates than larger dopants (such as Nb). This results in weaker bonds and higher overpotentials.


16

Page 17 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Conclusions We have created a model amorphous TiO2 surface that matches experimental data and shown that it contains reaction sites that predict ORR overpotentials in good agreement with experimentally measured values. Our computational models correctly predicted dopant trends and successfully identified dopants that were experimentally validated to lower ORR rates as much as –77 ± 3%. Optimizing the dopant concentration in the oxide can likely result in further ORR activity decreases. While we have demonstrated a straightforward method to predict dopants that modify metal oxides so that they more effectively suppress cathodic reduction kinetics, future work will address the ability of these doped Ti oxides to inhibit galvanic corrosion in realistic environments. Additionally, this doping approach should be able to increase the effectiveness of oxide protective coatings used on other materials/devices that suffer from corrosion damage (such as solar cell photoanodes42-43).

ASSOCIATED CONTENT Supporting Information. The Supporting information is available free of charge on the ACS Publications website. Computational model and experimental details (PDF)

Corresponding Author E-mail: [email protected], [email protected]


17


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 21

Notes The authors declare no competing financial interests. ACKNOWLEDGMENT This research was supported by the Office of Naval Research (grant number N0001414WX00918). The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the Office of Naval Research, the U.S. Navy or the U.S. government. This research was performed while RMA held an NRC Research Associateship award at the U.S. Naval Research Laboratory.

REFERENCES 1. Koch, G. H.; Varney, J.; Thompson, N. G.; Moghissi, O.; Gould, M.; Payer, J., International Measures of Prevention, Application, and Economics of Corrosion Technologies Study. NACE International: NACE International, Houston, Texas, USA, 2016; pp 2-6. 2. Donatus, U.; Thompson, G. E.; Zhou, X. R. Effect of Near-Ambient Temperature Changes on the Galvanic Corrosion of an AA2024-T3 and Mild Steel Couple. J. Electrochem. Soc. 2015, 162, C42-C46. 3. Andreatta, F.; Terryn, H.; de Wit, J. H. W. Effect of Solution Heat Treatment on Galvanic Coupling between Intermetallics and Matrix in AA7075-T6. Corros. Sci. 2003, 45, 1733-1746. 4. Sathiyanarayanan, S.; Muthukrishnan, S.; Venkatachari, G.; Trivedi, D. C. Corrosion Protection of Steel by Polyaniline (PANI) Pigmented Paint Coating. Prog. Org. Coat. 2005, 53, 297-301. 5. Samui, A. B.; Patankar, A. S.; Rangarajan, J.; Deb, P. C. Study of Polyaniline Containing Paint for Corrosion Prevention. Prog. Org. Coat. 2003, 47, 1-7. 6. De Baere, K.; Verstraelen, H.; Lemmens, L.; Lenaerts, S.; Dewil, R.; Van Ingelgem, Y.; Potters, G. A Field Study of the Effectiveness of Sacrificial Anodes in Ballast Tanks of Merchant Ships. J. Mar. Sci. Technol. 2014, 19, 116-123. 7. Lo, K. H.; Shek, C. H.; Lai, J. K. L. Recent Developments in Stainless Steels. Mater. Sci. Eng., R 2009, 65, 39-104. 8. Song, G. L. Recent Progress in Corrosion and Protection of Magnesium Alloys. Adv. Eng. Mater. 2005, 7, 563-586. 9. Mansfeld, F.; Parry, E. P. Galvanic Corrosion of Bare and Coated Al Alloys Coupled to Stainless Steel 304 or Ti-6Al-4V. Corros. Sci. 1973, 13, 605-621.


18

Page 19 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


10. Zhou, X.; Luo, C.; Hashimoto, T.; Hughes, A. E.; Thompson, G. E. Study of Localized Corrosion in AA2024 Aluminium Alloy Using Electron Tomography. Corros. Sci. 2012, 58, 299-306. 11. Park, H.; Kim, K. Y.; Choi, W. Photoelectrochemical Approach for Metal Corrosion Prevention Using a Semiconductor Photoanode. J. Phys. Chem. B 2002, 106, 4775-4781. 12. Nørskov, J. K.; Rossmeisl, J.; Logadottir, A.; Lindqvist, L.; Kitchin, J. R.; Bligaard, T.; Jónsson, H. Origin of the Overpotential for Oxygen Reduction at a Fuel-Cell Cathode. J. Phys. Chem. B 2004, 108, 17886-17892. 13. Fabbri, E.; Habereder, A.; Waltar, K.; Kotz, R.; Schmidt, T. J. Developments and Perspectives of Oxide-Based Catalysts for the Oxygen Evolution Reaction. Catal. Sci. Tech. 2014, 4, 3800-3821. 14. Liao, P.; Keith, J. A.; Carter, E. A. Water Oxidation on Pure and Doped Hematite (0001) Surfaces: Prediction of Co and Ni as Effective Dopants for Electrocatalysis. J. Am. Chem. Soc. 2012, 134, 13296-13309. 15. Greeley, J.; Markovic, N. M. The Road from Animal Electricity to Green Energy: Combining Experiment and Theory in Electrocatalysis. Energy Environ. Sci. 2012, 5, 9246-9256. 16. Morales-Guio, C. G.; Stern, L.-A.; Hu, X. Nanostructured Hydrotreating Catalysts for Electrochemical Hydrogen Evolution. Chem. Soc. Rev. 2014, 43, 6555-6569. 17. Kresse, G.; Furthmüller, J. Efficient Iterative Schemes for Ab Initio Total-Energy Calculations Using a Plane-Wave Basis Set. Phys. Rev. B 1996, 54, 11169-11186. 18. Kresse, G.; Furthmiiller, J. Efficiency of Ab-Initio Total Energy Calculations for Metals and Semiconductors Using a Plane-Wave Basis Set. Comput. Mater. Sci. 1996, 6, 15-50. 19. Kresse, G.; Hafner, J. Ab Initio Molecular-Dynamics Simulation of the Liquid-Metal Amorphous-Semiconductor Transition in Germanium. Phys. Rev. B 1994, 49, 14251-14269. 20. Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1997, 78, 1396-1396. 21. Kresse, G.; Joubert, D. From Ultrasoft Pseudopotentials to the Projector Augmented-Wave Method. Phys. Rev. B 1999, 59, 1758-1775. 22. Valdés, Á.; Qu, Z. W.; Kroes, G. J.; Rossmeisl, J.; Nørskov, J. K. Oxidation and PhotoOxidation of Water on TiO2 Surface. J. Phys. Chem. C 2008, 112, 9872-9879. 23. Mathew, K.; Sundararaman, R.; Letchworth-Weaver, K.; Arias, T. A.; Hennig, R. G. Implicit Solvation Model for Density-Functional Study of Nanocrystal Surfaces and Reaction Pathways. J. Chem. Phys. 2014, 140, 084106/1-8. 24. Mathew, K.; Hennig, R. G. Implicit Self-Consistent Description of Electrolyte in Plane-Wave Density-Functional Theory. https://arxiv.org/abs/1601.03346 (accessed November 30, 2016). 25. Krukau, A. V.; Vydrov, O. A.; Izmaylov, A. F.; Scuseria, G. E. Influence of the Exchange Screening Parameter on the Performance of Screened Hybrid Functionals. J. Chem. Phys. 2006, 125, 224106/1-5. 26. Aktulga, H. M.; Fogarty, J. C.; Pandit, S. A.; Grama, A. Y. Parallel Reactive Molecular Dynamics: Numerical Methods and Algorithmic Techniques. Parallel Comput. 2012, 38, 245259. 27. van Duin, A. C. T.; Dasgupta, S.; Lorant, F.; Goddard, W. A. ReaxFF: A Reactive Force Field for Hydrocarbons. J. Phys. Chem. A 2001, 105, 9396-9409. 28. Plimpton, S. Fast Parallel Algorithms for Short-Range Molecular Dynamics. J. Comput. Phys. 1995, 117, 1-19.


19


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 21

29. Kim, S.-Y.; Kumar, N.; Persson, P.; Sofo, J.; van Duin, A. C. T.; Kubicki, J. D. Development of a ReaxFF Reactive Force Field for Titanium Dioxide/Water Systems. Langmuir 2013, 29, 7838-7846. 30. Hangarter, C. M.; Policastro, S. A. Electrochemical Characterization of Galvanic Couples Under Saline Droplets in a Simulated Atmospheric Environment. Corrosion 2017, 73, 268-280. 31. Policastro, S. A.; Hangarter, C. M.; Horton, D. J.; Wollmershauser, J. A.; Roeper, D. F. Effect of Low-Concentration Alloying in Titanium on Reduction Reaction Kinetics in Alkaline Environments. J. Electrochem. Soc. 2016, 163, C269-C274. 32. Ewing, C. S.; Bhavsar, S.; Veser, G.; McCarthy, J. J.; Johnson, J. K. Accurate Amorphous Silica Surface Models from First-Principles Thermodynamics of Surface Dehydroxylation. Langmuir 2014, 30, 5133-5141. 33. Petkov, V.; Holzhüter, G.; Tröge, U.; Gerber, T.; Himmel, B. Atomic-Scale Structure of Amorphous TiO2 by Electron, X-ray Diffraction and Reverse Monte Carlo Simulations. J. NonCryst. Solids 1998, 231, 17-30. 34. Wang, Y.; Cheng, H.-P. Oxygen Reduction Activity on Perovskite Oxide Surfaces: A Comparative First-Principles Study of LaMnO3, LaFeO3, and LaCrO3. J. Phys. Chem. C 2013, 117, 2106-2112. 35. Su, H.-Y.; Gorlin, Y.; Man, I. C.; Calle-Vallejo, F.; Norskov, J. K.; Jaramillo, T. F.; Rossmeisl, J. Identifying Active Surface Phases for Metal Oxide Electrocatalysts: A Study of Manganese Oxide Bi-Functional Catalysts for Oxygen Reduction and Water Oxidation Catalysis. Phys. Chem. Chem. Phys. 2012, 14, 14010-14022. 36. Arashi, T.; Seo, J.; Takanabe, K.; Kubota, J.; Domen, K. Nb-Doped TiO2 Cathode Catalysts for Oxygen Reduction Reaction of Polymer Electrolyte Fuel Cells. Catal. Today 2014, 233, 181186. 37. Calle-Vallejo, F.; Loffreda, D.; Koper, M. T. M.; Sautet, P. Introducing Structural Sensitivity into Adsorption-Energy Scaling Relations by Means of Coordination Numbers. Nat. Chem. 2015, 7, 403-410. 38. Fung, V.; Tao, F. F.; Jiang, D.-e. General Structure–Reactivity Relationship for Oxygen on Transition-Metal Oxides. J. Phys. Chem. Lett. 2017, 8, 2206-2211. 39. Takeno, N., Atlas of Eh-pH Diagrams: Intercomparison of Thermodynamic Databases. In Geological Survey of Japan Open File Report No.419, National Institute of Advanced Industrial Science and Technology: 2005; pp 1-102. 40. Mehta, P.; Greeley, J.; Delgass, W. N.; Schneider, W. F. Adsorption Energy Correlations at the Metal–Support Boundary. ACS Catal. 2017, 4707-4715. 41. Lee, W.; Han, J. W.; Chen, Y.; Cai, Z. H.; Yildiz, B. Cation Size Mismatch and Charge Interactions Drive Dopant Segregation at the Surfaces of Manganite Perovskites. J. Am. Chem. Soc. 2013, 135, 7909-7925. 42. Hu, H.; Dong, B.; Hu, H.; Chen, F.; Kong, M.; Zhang, Q.; Luo, T.; Zhao, L.; Guo, Z.; Li, J.; Xu, Z.; Wang, S.; Eder, D.; Wan, L. Atomic Layer Deposition of TiO2 for a High-Efficiency Hole-Blocking Layer in Hole-Conductor-Free Perovskite Solar Cells Processed in Ambient Air. ACS Appl. Mater. Interfaces 2016, 8, 17999-18007. 43. Durand, S. J. Attaining a 30-Year Photovoltaic System Lifetime: The BOS Issues. Prog. Photovoltaics 1994, 2, 107-113.


20

Page 21 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


TOC Graphic


21

Doped Amorphous Ti Oxides To Deoptimize ... - ACS Publications

Recommend Documents